A method, system and medium for generating and standardizing three-dimensional morphology parameters of posterior pole

Abstract

The application provides a post-polar three-dimensional shape parameter generation and standardization method and system based on wide-field optical coherence tomography, and relates to the technical field of ophthalmic image analysis. In the prior art, when the post-polar OCT image shape parameter is extracted, the background interference caused by the macroscopic physiological curvature difference of the individual eyeball cannot be effectively stripped, the coordinate system lacks a unified anatomical reference, and the extracted shape parameter dimension is single. The application constructs a coordinate system based on the spatial positions of the optic disc center and the macular center, generates a reference surface through polynomial surface fitting and calculates residual height values to strip the overall shape trend, extracts multi-dimensional geometric features including slope, average curvature and Gaussian curvature based on the residual height field and performs parameterization characterization, and performs scale normalization and statistical standardization processing, thereby improving the separability of the local shape change characteristics and the accuracy of the post-polar shape parameters.

Description

Technical Field

[0001] This invention relates to the field of ophthalmic image analysis technology, and in particular to a method, system and medium for generating and standardizing three-dimensional morphological parameters of the posterior pole. Background Technology

[0002] With the rapid development of ophthalmic imaging technology, optical coherence tomography (OCT) has become an indispensable non-invasive imaging tool for assessing the posterior pole structure of the eyeball, diagnosing and monitoring various fundus diseases such as glaucoma, high myopia macular degeneration, and age-related macular degeneration. It allows clinicians to obtain a wider range and higher resolution of three-dimensional anatomical information covering the optic disc, macula, and peripheral regions in a single scan. This has crucial clinical value and broad application prospects for building large-scale clinical research databases, conducting multi-center clinical trials, and achieving precision ophthalmic diagnosis and treatment.

[0003] However, existing methods for extracting morphological parameters from posterior pole OCT images still face several challenges in practical clinical applications: First, current methods often rely directly on surface height data obtained from raw OCT scans for analysis, failing to effectively eliminate background interference caused by individual differences in the macroscopic physiological curvature of the eyeball. Second, the coordinate system construction of existing parameter extraction schemes lacks a unified benchmark. This leads to significant systematic biases in parameters acquired at different scanning time points or across different imaging devices, making it difficult to meet the stringent requirements for data comparability in longitudinal clinical follow-ups. Third, the morphological parameters extracted by existing technologies are often one-dimensional, failing to fully utilize the rich three-dimensional spatial information provided by wide-field OCT to perform multi-dimensional and refined mathematical characterization of local slope changes, surface curvature characteristics, and morphological asymmetry in the posterior pole, thus limiting their sensitivity in detecting early or atypical structural damage.

[0004] Therefore, there is an urgent need for a robust and standardized method for generating the three-dimensional morphological parameters of the rear pole that can overcome the above-mentioned defects. Summary of the Invention

[0005] To address these issues, this invention provides a method, system, and medium for generating and standardizing three-dimensional morphological parameters of the posterior pole, thereby overcoming the problems in existing technologies such as the failure to effectively isolate background interference caused by differences in the macroscopic physiological curvature of the eyeball between individuals, the lack of a unified benchmark in the coordinate system construction of existing parameter extraction schemes, and the fact that the morphological parameters extracted by existing technologies are often one-dimensional and fail to fully utilize the rich three-dimensional spatial information provided by wide-area OCT.

[0006] To achieve the above objectives, the present invention provides a method for generating and standardizing three-dimensional morphological parameters of the rear pole, comprising: S1, acquire wide-area optical coherence tomography image data of the subject's eyeball; S2, perform three-dimensional volume data reconstruction on the wide-area optical coherence tomography image data to obtain height distribution information characterizing the three-dimensional morphology of the posterior pole surface of the eyeball. S3, based on the spatial relationship between the center of the optic disc and the center of the macula, constructs an analytical reference coordinate system for the posterior pole of the eyeball; S4. Under the analysis reference coordinate system, determine the rear pole analysis region and perform meshing processing on the rear pole analysis region to generate meshed surface data composed of multiple mesh nodes, wherein each mesh node is associated with a three-dimensional spatial position. S5, perform smoothing filtering on the gridded surface data, and generate a reference surface representing the overall morphological trend of the posterior pole of the eyeball based on the smoothed gridded surface data. S6, calculate the residual height value of each grid node, wherein the residual height value is the difference between the original height value of the grid node and the corresponding height value of the reference surface at the grid node; S7. Based on the residual height value and the spatial distribution of the grid nodes, calculate the geometric features of each grid node in the post-polar analysis area; S8. Based on the geometric features and residual height value, the three-dimensional structural morphology in the analysis area of ​​the rear pole is parametrically characterized to generate rear pole morphological parameters that describe the degree of rear pole concavity, morphological asymmetry, and local curvature change characteristics. S9, the rear electrode morphological parameters are standardized to generate a standardized set of rear electrode morphological parameters that are comparable across devices and time points.

[0007] Furthermore, the geometric features include at least first-order geometric features and / or second-order geometric features, wherein the first-order geometric features include slope, and the second-order geometric features include at least one of mean curvature and Gaussian curvature.

[0008] Furthermore, the step of reconstructing three-dimensional volume data from the wide-area optical coherence tomography image data to obtain height distribution information characterizing the three-dimensional morphology of the posterior pole surface of the eyeball includes: S21, perform image preprocessing on the wide-area optical coherence tomography image data, the image preprocessing including at least one of noise suppression, image enhancement and image registration; S22, perform image segmentation on the image data after image preprocessing to extract the boundary contour of the Bruch membrane; S23, Based on the boundary contour, a continuous three-dimensional surface mesh covering the rear polar region is generated using a three-dimensional spatial interpolation algorithm; S24, the continuous three-dimensional surface mesh is converted into a height mapping map defined on a two-dimensional transverse plane coordinate domain, which serves as height distribution information characterizing the three-dimensional morphology of the posterior pole surface of the eyeball.

[0009] Furthermore, the step of determining the rear pole analysis region in the analysis reference coordinate system and performing meshing processing on the rear pole analysis region to generate meshed surface data composed of multiple mesh nodes includes: S41, In the analysis reference coordinate system, the line connecting the center of the optic disc and the center of the macula is used as the reference baseline. Based on the preset extension range, the posterior pole analysis area covering the area around the optic disc and the area around the macula is determined. S42, on the two-dimensional transverse plane projection of the rear pole analysis region, a two-dimensional planar mesh is generated according to the preset mesh resolution; S43, Based on the height distribution information, determine the height value corresponding to each grid node in the two-dimensional planar grid; S44 combines the height value of each grid node with the two-dimensional planar grid to construct a grid node with a three-dimensional spatial position.

[0010] Furthermore, the step of smoothing and filtering the gridded surface data, and then fitting and generating a reference surface representing the overall morphological trend of the posterior pole of the eyeball based on the smoothed and filtered gridded surface data, includes: S51, Spatial domain smoothing filtering is performed on the height values ​​of each grid node in the gridded surface data to eliminate high-frequency noise and local small fluctuations, so as to obtain smooth gridded surface data. S52, Based on the smooth gridded surface data, a low-order polynomial surface representing the overall bending trend of the posterior pole of the eyeball is obtained through a polynomial surface fitting algorithm. S53, the low-order polynomial surface is used as the reference surface to define a smooth reference surface that reflects the macroscopic anatomical curvature of the posterior pole in the analysis reference coordinate system.

[0011] Furthermore, the calculation of the geometric features at each grid node within the post-polarity analysis region based on the residual height value and the spatial distribution of the grid nodes includes: S71, Based on the two-dimensional plane coordinates of each grid node in the gridded surface data and the corresponding residual height value, construct a residual height field function with two-dimensional plane coordinates as independent variables and residual height value as dependent variables; S72, on the residual height field function, based on the residual height values ​​of each grid node and its neighboring nodes, the first-order partial derivative and / or second-order partial derivative at each network node are calculated by numerical differentiation method. S73, calculate the slope at each network node based on the first-order partial derivative, and / or calculate the average curvature and / or Gaussian curvature at each network node based on the first-order and second-order partial derivatives, as geometric features.

[0012] Furthermore, the parameterization is achieved by extracting quantitative indicators through statistical analysis methods and combining the quantitative indicators into a set of parameter vectors as morphological parameters of the posterior pole. The quantitative indicators include at least a global statistic reflecting the overall degree of concavity of the posterior pole, a spatial distribution difference reflecting morphological asymmetry, and a regional extreme value or gradient statistic reflecting the characteristics of local curvature changes.

[0013] Furthermore, the standardization of the rear electrode morphological parameters to generate a standardized set of rear electrode morphological parameters with cross-device and cross-time point comparability includes: S91: Obtain the original dimensions and numerical range of each parameter in the morphological parameters of the rear pole; S92: According to the preset standardization rules, the morphological parameters of the rear pole are normalized in terms of scale. S93: The posterior pole morphological parameters after scale normalization are compared with the corresponding statistical distribution characteristics in the preset reference population database, and statistical standardization is performed to obtain standardized parameter values. S94: Combine the standardized parameter values ​​to generate the standardized polarity parameter set.

[0014] This invention also provides a system for generating and standardizing three-dimensional morphological parameters of the posterior pole, the system being used to implement any of the methods for generating and standardizing three-dimensional morphological parameters of the posterior pole, the system comprising: The image acquisition module is used to acquire wide-area optical coherence tomography image data of the subject's eyeball; The three-dimensional reconstruction module, connected to the image acquisition module, is used to perform three-dimensional volume data reconstruction on the wide-area optical coherence tomography image data to obtain height distribution information characterizing the three-dimensional morphology of the posterior pole surface of the eyeball. The coordinate system construction module, connected to the three-dimensional reconstruction module, is used to construct an analytical reference coordinate system for the posterior pole of the eyeball based on the spatial relationship between the center of the optic disc and the center of the macula. The meshing module, connected to the coordinate system construction module, is used to determine the rear pole analysis region under the analysis reference coordinate system and perform meshing processing on the rear pole analysis region to generate meshed surface data composed of multiple mesh nodes, wherein each mesh node is associated with a three-dimensional spatial position. The reference surface fitting module, connected to the meshing processing module, is used to perform smoothing filtering on the meshed surface data, and to fit and generate a reference surface representing the overall morphological trend of the posterior pole of the eyeball based on the smoothed meshed surface data. The residual calculation module, connected to the reference surface fitting module, is used to calculate the residual height value of each grid node. The residual height value is the difference between the original height value of the grid node and the corresponding height value of the reference surface at the grid node. The geometric feature calculation module, connected to the residual calculation module, is used to calculate the geometric features at each grid node in the post-polar analysis area based on the residual height value and the spatial distribution of grid nodes. The parametric characterization module, connected to the geometric feature calculation module, is used to parametrically characterize the three-dimensional structural morphology of the rear pole analysis area based on the geometric features and residual height value, and generate rear pole morphological parameters to describe the degree of rear pole concavity, morphological asymmetry and local curvature change characteristics. The parameter standardization module, connected to the parameterization characterization module, is used to standardize the morphological parameters of the rear pole to generate a standardized set of morphological parameters of the rear pole that is comparable across devices and time points.

[0015] The present invention also provides a storage medium for computer-readable storage, wherein the storage medium stores one or more programs that can be executed by one or more processors to implement the method for generating and standardizing the three-dimensional morphological parameters of the rear pole as described in any one of the claims.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows: Firstly, this invention solves the problem in the prior art where background interference caused by the failure to effectively remove the differences in macroscopic physiological curvature of the eyeball between individuals leads to the obscuring of focal lesion features by the overall curved shape. This is achieved by employing a technique of smoothing and filtering the gridded curved surface data of the posterior pole and fitting it to generate a reference curved surface that characterizes the overall morphological trend of the posterior pole of the eyeball. Then, the difference between the original height value of each grid node and the corresponding height value of the reference curved surface is calculated as the residual height value. This improves the separability of focal morphological change features from the overall background. Secondly, this invention solves the problems of lack of a unified anatomical reference coordinate system in the construction of the coordinate system and systematic deviation of morphological parameters caused by the difference in scanning parameters of different devices and different time points in the prior art by adopting a unified anatomical analysis reference coordinate system based on the spatial positional relationship between the optic disc center and the macula center, and by performing scale normalization and statistical standardization processing on the morphological parameters of the posterior pole by combining preset standardization rules. This improves the comparability and clinical applicability of the morphological parameters of the posterior pole across devices and time points. Thirdly, this invention integrates benchmark surface fitting and residual separation techniques, unified coordinate system and parameter standardization techniques, and multidimensional geometric feature calculation and parameterized characterization techniques based on residual height fields to construct a complete technical solution from wide-area OCT image acquisition to standardized morphological parameter set generation. This improves the anatomical consistency, feature description precision and multidimensionality of posterior pole three-dimensional morphological analysis, as well as the universality and comparability of clinical applications, effectively overcoming the shortcomings of existing technologies such as large background interference, inconsistent benchmarks, and single parameter dimensions. Attached Figure Description

[0017] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0018] Figure 1 A flowchart illustrating a method for generating and standardizing three-dimensional morphological parameters of the rear pole portion, provided in an embodiment of the present invention; Figure 2 This is a structural block diagram of a system for generating and standardizing three-dimensional morphological parameters of the rear pole, provided in an embodiment of the present invention. Detailed Implementation

[0019] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0020] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0021] It should be noted that in the description of this invention, the terms "upper", "lower", "left", "right", "inner", "outer", etc., which indicate directions or positional relationships, are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and is not intended to indicate or imply that the device or element must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of this invention.

[0022] Furthermore, it should be noted that, in the description of this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0023] Example 1 like Figure 1 As shown, this invention provides a method for generating and standardizing three-dimensional morphological parameters of the rear pole, including: S1, acquire wide-area optical coherence tomography image data of the subject's eyeball; S2, perform three-dimensional volume data reconstruction on the wide-area optical coherence tomography image data to obtain height distribution information characterizing the three-dimensional morphology of the posterior pole surface of the eyeball, including: S21, perform image preprocessing on the wide-area optical coherence tomography image data, the image preprocessing including at least one of noise suppression, image enhancement and image registration; S22, perform image segmentation on the image data after image preprocessing to extract the boundary contour of the Bruch membrane; S23, Based on the boundary contour, a continuous three-dimensional surface mesh covering the rear polar region is generated using a three-dimensional spatial interpolation algorithm; S24, the continuous three-dimensional surface mesh is converted into a height mapping map defined on a two-dimensional transverse plane coordinate domain, which serves as height distribution information characterizing the three-dimensional morphology of the posterior pole surface of the eyeball.

[0024] In one possible implementation, wide-area optical coherence tomography (OCT) images of the subject's eyeball are acquired using a commercially available wide-area OCT device. The wide-area OCT scanning protocol is configured to scan a rectangular area of ​​12mm × 9mm centered on the fovea of ​​the macula, containing multiple B-scan tomographic sections, each B-scan consisting of a certain number of A-scan scan lines. The acquired image data is stored in a standard medical image format, such as DICOM, which includes information on the light reflection intensity of the posterior pole three-dimensional structure.

[0025] Image preprocessing is performed on the acquired image data to improve the accuracy and stability of subsequent segmentation and reconstruction. Image preprocessing may include one or more of the following: noise suppression, image enhancement, and image registration. Specifically, a three-dimensional Gaussian filter or a three-dimensional median filter is used to smooth the original volume data to suppress speckle noise inherent in OCT imaging; a contrast-limited adaptive histogram equalization algorithm is used to enhance the contrast of each layer of the retina, making the target boundary clearer; if the wide-area OCT image is stitched together from multiple sub-regions, image registration of each sub-region is required. Optionally, a rigid registration algorithm based on mutual information measure is used to align each sub-region to the same spatial coordinate system, eliminating stitching artifacts caused by factors such as eye movement.

[0026] Image segmentation was performed on the preprocessed image data to extract the boundary contour of the Bruch's membrane. The Bruch's membrane is a thin layer located between the retinal pigment epithelium and the choroidal capillary layer. In OCT images, it appears as a highly reflective, continuous curve and is an important anatomical landmark for defining the outer boundary of the retina and assessing the morphology of the posterior pole. Specifically, graph-based image segmentation algorithms, such as Graph Cut or random walk algorithms, or deep learning-based semantic segmentation networks, were used to segment each B-scan tomographic image layer by layer, automatically identifying and extracting the positional coordinate sequence of the Bruch's membrane in the two-dimensional tomographic plane. The Bruch's membrane contour lines from all B-scans were then superimposed in three-dimensional space to obtain discrete three-dimensional point cloud data of the Bruch's membrane in the posterior pole region.

[0027] Radial basis function interpolation or Kriging interpolation is used to fit spatial surfaces to discrete point cloud data. Specifically, the Bruch membrane point cloud data can be projected onto a two-dimensional planar domain, and the height value, i.e., the Z-axis coordinate perpendicular to this plane, is interpolated to generate a continuous, void-free height field function within the entire aft pole region of interest defined by the visual disk and macula. Based on this, the Delaunay triangulation algorithm or regular mesh partitioning algorithm can be used to generate a continuous three-dimensional surface mesh composed of multiple triangular or quadrilateral facets.

[0028] A two-dimensional transverse plane coordinate system is established with the macula center as the origin and the line connecting the optic disc center and the macula center as a coordinate axis. For any coordinate point (x, y) on this plane, its height value z in the height mapping map is obtained by looking up the height value of the continuous surface generated in step S23 at the corresponding position through table lookup or linear interpolation. Thus, the three-dimensional morphology of the posterior pole surface is completely expressed as a single-valued function z = f(x, y). This invention addresses the problems of inaccurate posterior pole morphological representation and poor repeatability caused by unclear segmentation targets or the selection of suboptimal anatomical landmarks in existing technologies. By segmenting wide-area OCT image data to specifically extract the Bruch membrane boundary contour, it solves these problems. Using the Bruch membrane as the basic boundary for 3D reconstruction most realistically reflects the macroscopic structural morphology of the sclera in the posterior pole of the eyeball, improving the accuracy and repeatability of posterior pole height distribution information extraction. Furthermore, by using 3D spatial interpolation based on the discrete boundary contour point cloud of the Bruch membrane to generate a continuous 3D surface mesh, it solves the problems of insufficient spatial sampling, surface discontinuities, and susceptibility to local scan gaps when directly estimating 3D morphology from discrete B-scan tomographic lines in existing technologies. It reconstructs a mathematically smooth and continuous posterior pole surface morphology covering the entire posterior pole analysis area from limited discrete point cloud data. This method improves the completeness and robustness of the overall curvature morphology estimation of the rear pole, while reducing the error caused by local defects or noise interference in the original scan data on the final three-dimensional morphology reconstruction result. By further converting the continuous three-dimensional surface mesh into a height mapping map defined on the two-dimensional transverse plane coordinate domain, it solves the problems of inconsistent three-dimensional surface data formats and high computational complexity of subsequent meshing and differential geometric features, reducing computational complexity and algorithm implementation difficulty, and improving the execution efficiency and feasibility of the method.

[0029] S3, based on the spatial relationship between the center of the optic disc and the center of the macula, constructs an analytical reference coordinate system for the posterior pole of the eyeball; S4, in the analysis reference coordinate system, the rear pole analysis region is determined, and the rear pole analysis region is meshed to generate meshed surface data composed of multiple mesh nodes, wherein each mesh node is associated with a three-dimensional spatial position, including: S41, In the analysis reference coordinate system, the line connecting the center of the optic disc and the center of the macula is used as the reference baseline. Based on the preset extension range, the posterior pole analysis area covering the area around the optic disc and the area around the macula is determined. S42, on the two-dimensional transverse plane projection of the rear pole analysis region, a two-dimensional planar mesh is generated according to the preset mesh resolution; S43, Based on the height distribution information, determine the height value corresponding to each grid node in the two-dimensional planar grid; S44 combines the height value of each grid node with the two-dimensional planar grid to construct a grid node with a three-dimensional spatial position.

[0030] In one possible implementation, the spatial locations of the optic disc center and the macula center are automatically or semi-automatically identified in the height map or raw OCT image data. Optionally, the location of the optic disc center can be achieved by detecting the position of the Bruch's membrane opening at the edge of the optic disc in the OCT image, or by using a deep learning-based object detection network to identify the optic disc region and calculate its geometric center. The location of the macula center can be automatically determined by searching for local minima in the height distribution information, utilizing the anatomical features of the fovea where the retina is thinnest and the Bruch's membrane height is lowest.

[0031] After obtaining the two-dimensional planar coordinates of the optic disc center and the macula center, an analytical reference coordinate system is constructed. Optionally, the macula center is taken as the origin of the coordinate system, the vector direction from the macula center to the optic disc center is taken as the positive X-axis direction, and the direction perpendicular to the X-axis in the two-dimensional transverse plane is taken as the positive Y-axis direction.

[0032] In this embodiment, the preset extension range is 3.0 mm from the center of the macula towards the temporal side and 4.0 mm towards the nasal side along the X-axis, so that the total span of the analysis area in the X-axis direction covers the main structural areas of the macula and optic disc; along the Y-axis direction, it extends 3.5 mm upwards and downwards from the X-axis, forming a rectangular analysis area. The preset grid resolution is 50 μm × 50 μm, that is, a grid node is taken every 50 μm in the X-axis and Y-axis directions. Thus, within the rectangular analysis area of ​​the example above, several regular rectangular grids are generated, where each node corresponds to a unique two-dimensional planar coordinate.

[0033] Since the surface morphology of the rear pole has been expressed as a height map defined on a two-dimensional transverse plane coordinate domain, the height value of any grid node can be directly obtained by querying or interpolating within this height map. If a grid node falls exactly on a known sampling point of the height map, the height value of that point is directly read; if the grid node is located between sampling points, bilinear or bicubic interpolation algorithms can be used to calculate the height value of the grid node based on the height values ​​of neighboring sampling points. Combining the height value of each grid node with its corresponding two-dimensional plane coordinates constructs grid nodes with three-dimensional spatial positions, forming the gridded surface data characterizing the three-dimensional morphology of the rear pole surface within the analysis region.

[0034] This invention addresses the problems of inconsistent coordinate system definitions and unstable spatial correspondences across individuals and time points caused by relying solely on single anatomical landmarks or manual calibration for image alignment in existing technologies. It employs a technique to construct an analytical reference coordinate system for the posterior pole of the eye based on the spatial relationship between the optic disc center and the macula center. This improves the spatial consistency and accuracy of morphological parameter extraction. Furthermore, by performing regular meshing and assigning height values ​​to each node based on height distribution information to construct a set of mesh nodes with three-dimensional spatial positions, this invention solves the problems of inconsistent data representation formats and unclear node spatial topological relationships in existing technologies, which lead to difficulties or insufficient accuracy in subsequent differential geometric feature calculations. This improves the spatial topological accuracy of the posterior pole surface data structure.

[0035] S5, perform smoothing filtering on the gridded surface data, and generate a reference surface representing the overall morphological trend of the posterior pole of the eyeball based on the smoothed gridded surface data, including: S51, Spatial domain smoothing filtering is performed on the height values ​​of each grid node in the gridded surface data to eliminate high-frequency noise and local small fluctuations, so as to obtain smooth gridded surface data. S52, Based on the smooth gridded surface data, a low-order polynomial surface representing the overall bending trend of the posterior pole of the eyeball is obtained through a polynomial surface fitting algorithm. S53, the low-order polynomial surface is used as the reference surface to define a smooth reference surface that reflects the macroscopic anatomical curvature of the posterior pole in the analysis reference coordinate system.

[0036] S6, calculate the residual height value of each grid node, wherein the residual height value is the difference between the original height value of the grid node and the corresponding height value of the reference surface at the grid node; In one possible implementation, due to the inherent speckle noise during OCT imaging and the slight fluctuations that may be introduced by the 3D interpolation in step S23, the meshed surface data generated in step S4 may contain high-frequency components that do not reflect the true anatomical structure. If these high-frequency components are not filtered out, they will cause overfitting in the polynomial surface fitting in subsequent steps, affecting the representation effect of the reference surface on the overall morphological trend. In this embodiment, a two-dimensional Gaussian filter is used to convolve the height value matrix, and the standard deviation σ of the Gaussian filter kernel is set to a value in the range of 0.3 mm to 0.5 mm, and the filter kernel window size is set to a rectangular area of ​​3σ×3σ. This parameter setting can effectively filter out high-frequency noise and slight fluctuations while retaining the macroscopic bending morphology of the posterior pole and the mesoscale structural change information.

[0037] The macroscopic anatomical curvature of the posterior pole of the eyeball can typically be mathematically described by a smooth, continuous low-order surface. Therefore, this step employs a polynomial surface equation to parametrically model this macroscopic morphology. In a specific example, a second-order polynomial surface equation is used to perform least-squares fitting on the smoothed mesh node height values. The mathematical expression of the second-order polynomial surface equation is: ; Where (x, y) are the two-dimensional planar coordinates of the mesh nodes in the analysis reference coordinate system. Let a0 be the height value of the reference surface at this coordinate, and a5 be the fitting coefficients to be determined. Substituting all grid node data into the above equation, an overdetermined linear equation system is constructed. Solving this equation system using the least squares method yields the optimal coefficient vector that minimizes the sum of squared fitting residuals, thus obtaining a uniquely determined second-order polynomial surface equation. For any two-dimensional plane coordinate within the analysis region, its corresponding height value on the reference surface can be directly calculated by substituting the coordinates into the polynomial equation. The reference surface represents the individualized overall curvature baseline of the posterior pole of the subject's eyeball.

[0038] After obtaining the reference surface, the residual height value of each grid node is calculated to isolate the overall morphological trend and highlight local structural changes. Specifically, for each grid node in the meshed surface data, the difference between the original height value and the corresponding height value of the reference surface at that node is the residual height value of that grid node. This is calculated for all grid nodes in the analysis area to obtain a residual height matrix. In this residual height matrix, a positive value indicates that the actual rear pole surface at that node is higher than the reference surface, i.e., it is locally raised relative to the overall bending shape, while a negative value indicates that the actual rear pole surface at that node is lower than the reference surface, i.e., it is locally concave relative to the overall bending shape.

[0039] This invention addresses the high-frequency artifact interference caused by OCT speckle noise and interpolation fluctuations in the original gridded surface data by using spatial domain smoothing filtering on the height values ​​of each grid node, thus improving the signal-to-noise ratio of the posterior pole surface height data. Furthermore, it solves the problem of local lesion features being masked by the overall morphology due to the failure of existing technologies to effectively separate the individual's macroscopic physiological curvature by using a polynomial surface fitting algorithm to obtain a low-order polynomial surface characterizing the overall curvature trend of the posterior pole of the eyeball, improving the accuracy of fitting the reference surface to the macroscopic anatomical curvature of the individual posterior pole. Finally, by defining the low-order polynomial surface as the reference surface and calculating the residual height values, it solves the problem of the overall morphological trend of the posterior pole and focal structural changes being intertwined in the height data and difficult to analyze independently, improving the separability of focal morphological features from the overall background.

[0040] S7, based on the residual height value and the spatial distribution of the grid nodes, calculate the geometric features at each grid node within the post-polarity analysis region, including: S71, Based on the two-dimensional plane coordinates of each grid node in the gridded surface data and the corresponding residual height value, construct a residual height field function with two-dimensional plane coordinates as independent variables and residual height value as dependent variables; S72, on the residual height field function, based on the residual height values ​​of each grid node and its neighboring nodes, the first-order partial derivative and / or second-order partial derivative at each network node are calculated by numerical differentiation method. S73, calculate the slope at each network node based on the first-order partial derivative, and / or calculate the average curvature and / or Gaussian curvature at each network node based on the first-order and second-order partial derivatives, as geometric features.

[0041] The geometric features include at least first-order geometric features and / or second-order geometric features, wherein the first-order geometric features include slope, and the second-order geometric features include at least one of mean curvature and Gaussian curvature.

[0042] In one possible implementation, the residual height values ​​of all grid nodes within the analysis region have been calculated. Each node corresponds to a two-dimensional plane coordinate. The discrete node data is organized into a continuous field function. For any given coordinate point, if the point is located exactly on a grid node, the function value is equal to the residual height value of that node. If the point is located between grid nodes, the function value can be calculated based on the residual height values ​​of neighboring nodes through bilinear interpolation or bicubic interpolation.

[0043] Slope is a first-order geometric feature characterizing the degree of inclination of a surface at a point. It is defined as the tangent of the angle between the tangent plane at that point and the horizontal plane, and can be directly calculated from the first-order partial derivative. A larger slope value indicates a more drastic change in residual height at that node; a smaller slope value indicates a more gradual change in residual height at that node. In posterior pole morphological analysis, slope can be used to identify the steepness of the optic disc margin or the boundary transition characteristics of focal lesions.

[0044] Mean curvature is a second-order geometric feature that characterizes the average degree of curvature of a surface at a given point. A positive mean curvature indicates that the surface at that point is locally convex, meaning it rises upwards relative to the surrounding area; a negative mean curvature indicates that the surface at that point is locally concave, meaning it collapses downwards relative to the surrounding area. In posterior pole morphological analysis, mean curvature can be used to quantify the depth and shape of optic disc cupping, or the degree of bulging of posterior staphyloma.

[0045] Gaussian curvature is a second-order geometric characteristic that characterizes the inherent curvature of a surface at a given point. The sign of the Gaussian curvature can be used to distinguish local surface morphology types: a positive value indicates that the surface at that point is locally elliptical, a negative value indicates that the surface at that point is locally hyperbolic, and a zero value indicates that the surface at that point is locally parabolic. In posterior pole morphology analysis, Gaussian curvature can be used to distinguish between different morphological types, such as the funnel-shaped depression in the optic disc region and the gentle depression in the macular region.

[0046] This invention addresses the problem in existing technologies that rely on a single dimensionality of morphological parameters and focus only on overall average values, failing to fully utilize wide-area OCT three-dimensional spatial information for a refined mathematical representation of the local geometry of the rear pole surface. This is achieved by employing a technique based on residual height values ​​and the spatial distribution of grid nodes to calculate the geometric features at each grid node within the post-pole analysis region. This technique significantly improves the precision and multidimensionality of the description of the local morphological features of the three-dimensional surface of the rear pole.

[0047] S8. Based on the geometric features and residual height value, the three-dimensional structural morphology in the analysis area of ​​the rear pole is parametrically characterized to generate rear pole morphological parameters that describe the degree of rear pole concavity, morphological asymmetry, and local curvature change characteristics. Parametric characterization involves extracting quantitative indicators through statistical analysis methods and combining these quantitative indicators into a set of parameter vectors as morphological parameters of the posterior pole. The quantitative indicators include at least a global statistic reflecting the overall degree of concavity of the posterior pole, a spatial distribution difference reflecting morphological asymmetry, and a regional extreme value or gradient statistic reflecting the characteristics of local curvature changes.

[0048] S9, standardize the rear electrode morphological parameters to generate a standardized set of rear electrode morphological parameters that is comparable across devices and time points, including: S91: Obtain the original dimensions and numerical range of each parameter in the morphological parameters of the rear pole; S92: According to the preset standardization rules, the morphological parameters of the rear pole are normalized in terms of scale. S93: The posterior pole morphological parameters after scale normalization are compared with the corresponding statistical distribution characteristics in the preset reference population database, and statistical standardization is performed to obtain standardized parameter values. S94: Combine the standardized parameter values ​​to generate the standardized polarity parameter set.

[0049] In one possible implementation, a statistic of the residual height values ​​of all grid nodes within the analysis region is used as a quantitative indicator reflecting the overall degree of concavity at the rear pole. In a specific example, one or more of the following statistics can be calculated: The average residual height is the arithmetic mean of the residual height values ​​of all grid nodes within the analysis area, reflecting the overall direction and extent of the offset of the rear polar surface relative to the reference surface. A negative value indicates an overall concave trend. The standard deviation of residual height reflects the overall dispersion of residual height values; Percentiles of specific regions, for example, the 5th percentile of residual height values ​​in the region surrounding the optic disc, are selected as an indicator of the depth of focal depression.

[0050] By comparing and analyzing the differences in morphological parameters between different spatial partitions within a region, quantitative indicators reflecting morphological asymmetry are extracted. In a specific example, one or more of the following differences are calculated: The asymmetry ratio between the upper and lower hemispheres is determined by dividing the analysis area into an upper and lower half, with the X-axis of the analysis reference coordinate system as the boundary. The average residual height or average curvature value of the upper and lower half is calculated respectively, and the ratio or difference between the two is used as the asymmetry index. The nasotemporal asymmetry ratio is calculated by dividing the analysis area into the nasal and temporal regions using a straight line passing through the center of the macula and perpendicular to the X-axis. The average slope or average Gaussian curvature value of each side is calculated, and the ratio or difference between the two is used as an indicator of asymmetry.

[0051] Based on the calculated geometric features, quantitative indicators reflecting local curvature variations are extracted. In a specific example, one or more of the following statistics can be calculated: The extreme value of the average curvature of the optic disc region is the minimum value of the average curvature in the area surrounding the optic disc, which serves as an indicator to quantify the steepness of the optic disc concavity. Curvature gradient: The gradient of the change in average curvature or Gaussian curvature along a specific direction within the analysis area is calculated as an indicator of the degree of drastic change in shape. Focal lesion area identification and quantification: By setting a curvature threshold, it automatically identifies local areas with abnormal curvature and extracts the area and average curvature value of the region.

[0052] One or more of the extracted quantitative indicators are combined into a parameter vector, which serves as the posterior pole morphological parameters. Due to differences in scanning resolution, signal strength, and data post-processing algorithms among different OCT devices, and the potential for variations in scanning parameters at different time points for the same subject, the generated raw posterior pole morphological parameters may exhibit systematic biases, making them difficult to directly use for comparisons across devices or time points. Therefore, the raw dimensions and numerical ranges of each parameter in the posterior pole morphological parameters are obtained. For example, the dimension of the average residual height parameter is length; the dimension of the asymmetry ratio parameter is a dimensionless ratio; and the dimension of the curvature parameter is the reciprocal of the length unit.

[0053] In one specific example, for parameters with length dimensions, individual eye-specific parameters such as axial length or corneal curvature radius can be used as normalization factors for scale correction. For instance, dividing the mean residual height value by the axial length yields a normalized relative concavity index. In another specific example, for curvature parameters, uniform correction can be performed based on the scanning resolution or image scaling factor of the OCT device. Optionally, the Z-score normalization method can also be used, which involves subtracting the mean from each parameter and then dividing by the standard deviation to convert it into a dimensionless standard score.

[0054] The reference population database is a pre-established database containing the statistical distribution characteristics of posterior pole morphological parameters of a specific population. In a specific example, this database may contain statistics such as the mean, standard deviation, and percentile distribution of various posterior pole morphological parameters of normal healthy individuals.

[0055] Specific methods of statistical standardization may include Z-score transformation and percentile mapping. The standardized parameter values ​​after statistical standardization are combined to form the standardized polar morphological parameter set. This standardized polar morphological parameter set eliminates systematic biases introduced by differences in equipment and scanning parameters through scale normalization, and maps individual parameter values ​​to a unified statistical framework of a reference population through statistical standardization. Therefore, this parameter set can be directly used for quantitative comparisons between image data acquired by different devices or between different follow-up time points of the same subject, providing comparable and interpretable standardized morphological indicators.

[0056] This invention addresses the shortcomings of existing technologies, such as the single-dimensional morphological parameter, focus on overall average values, and failure to provide a multi-dimensional and refined mathematical description of the posterior pole's three-dimensional surface. It employs a technique for parametrically representing the three-dimensional structural morphology of the posterior pole based on geometric features and residual height values. This improves the multidimensionality and structural completeness of the posterior pole morphological parameter representation. Furthermore, by normalizing the posterior pole morphological parameters according to preset standardization rules, it resolves the issues of differences in scanning resolution, signal intensity, and post-processing algorithms among different OCT devices, as well as the systematic bias in original morphological parameters caused by varying scanning parameters at different time points for the same subject. This enhances the numerical consistency of posterior pole morphological parameters across devices and time points. Finally, through statistical standardization, it addresses the lack of unified reference standards for individual morphological parameters and the difficulty in quantitatively assessing the degree of abnormality in clinical practice, improving the clinical universality and statistical interpretability of posterior pole morphological parameters.

[0057] Example 2 like Figure 2As shown, this invention provides a system for generating and standardizing three-dimensional morphological parameters of the posterior pole. The system is used to implement the method for generating and standardizing three-dimensional morphological parameters of the posterior pole as described in any of Embodiment 1. The system includes: The image acquisition module is used to acquire wide-area optical coherence tomography image data of the subject's eyeball; The three-dimensional reconstruction module, connected to the image acquisition module, is used to perform three-dimensional volume data reconstruction on the wide-area optical coherence tomography image data to obtain height distribution information characterizing the three-dimensional morphology of the posterior pole surface of the eyeball. The coordinate system construction module, connected to the three-dimensional reconstruction module, is used to construct an analytical reference coordinate system for the posterior pole of the eyeball based on the spatial relationship between the center of the optic disc and the center of the macula. The meshing module, connected to the coordinate system construction module, is used to determine the rear pole analysis region under the analysis reference coordinate system and perform meshing processing on the rear pole analysis region to generate meshed surface data composed of multiple mesh nodes, wherein each mesh node is associated with a three-dimensional spatial position. The reference surface fitting module, connected to the meshing processing module, is used to perform smoothing filtering on the meshed surface data, and to fit and generate a reference surface representing the overall morphological trend of the posterior pole of the eyeball based on the smoothed meshed surface data. The residual calculation module, connected to the reference surface fitting module, is used to calculate the residual height value of each grid node. The residual height value is the difference between the original height value of the grid node and the corresponding height value of the reference surface at the grid node. The geometric feature calculation module, connected to the residual calculation module, is used to calculate the geometric features at each grid node in the post-polar analysis area based on the residual height value and the spatial distribution of grid nodes. The parametric characterization module, connected to the geometric feature calculation module, is used to parametrically characterize the three-dimensional structural morphology of the rear pole analysis area based on the geometric features and residual height value, and generate rear pole morphological parameters to describe the degree of rear pole concavity, morphological asymmetry and local curvature change characteristics. The parameter standardization module, connected to the parameterization characterization module, is used to standardize the morphological parameters of the rear pole to generate a standardized set of morphological parameters of the rear pole that is comparable across devices and time points.

[0058] Example 3 A storage medium for computer-readable storage, wherein the storage medium stores one or more programs that can be executed by one or more processors to implement the method for generating and standardizing the three-dimensional morphological parameters of the posterior pole as described in any of Embodiment 1.

[0059] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A method for generating and standardizing three-dimensional morphological parameters of the rear pole, characterized in that, include: S1, acquire wide-area optical coherence tomography image data of the subject's eyeball; S2, perform three-dimensional volume data reconstruction on the wide-area optical coherence tomography image data to obtain height distribution information characterizing the three-dimensional morphology of the posterior pole surface of the eyeball. S3, based on the spatial relationship between the center of the optic disc and the center of the macula, constructs an analytical reference coordinate system for the posterior pole of the eyeball; S4. Under the analysis reference coordinate system, determine the rear pole analysis region and perform meshing processing on the rear pole analysis region to generate meshed surface data composed of multiple mesh nodes, wherein each mesh node is associated with a three-dimensional spatial position. S5, perform smoothing filtering on the gridded surface data, and generate a reference surface representing the overall morphological trend of the posterior pole of the eyeball based on the smoothed gridded surface data. S6, calculate the residual height value of each grid node, wherein the residual height value is the difference between the original height value of the grid node and the corresponding height value of the reference surface at the grid node; S7. Based on the residual height value and the spatial distribution of the grid nodes, calculate the geometric features of each grid node in the post-polar analysis area; S8. Based on the geometric features and residual height value, the three-dimensional structural morphology in the analysis area of ​​the rear pole is parametrically characterized to generate rear pole morphological parameters that describe the degree of rear pole concavity, morphological asymmetry, and local curvature change characteristics. S9, the rear electrode morphological parameters are standardized to generate a standardized set of rear electrode morphological parameters that are comparable across devices and time points.

2. The method for generating and standardizing the three-dimensional morphological parameters of the rear pole as described in claim 1, characterized in that, The geometric features include at least first-order geometric features and / or second-order geometric features, wherein the first-order geometric features include slope, and the second-order geometric features include at least one of mean curvature and Gaussian curvature.

3. The method for generating and standardizing the three-dimensional morphological parameters of the rear pole as described in claim 1, characterized in that, The step of reconstructing three-dimensional volume data from the wide-area optical coherence tomography image data to obtain height distribution information characterizing the three-dimensional morphology of the posterior pole surface of the eyeball includes: S21, perform image preprocessing on the wide-area optical coherence tomography image data, the image preprocessing including at least one of noise suppression, image enhancement and image registration; S22, perform image segmentation on the image data after image preprocessing to extract the boundary contour of the Bruch membrane; S23, Based on the boundary contour, a continuous three-dimensional surface mesh covering the rear polar region is generated using a three-dimensional spatial interpolation algorithm; S24, the continuous three-dimensional surface mesh is converted into a height mapping map defined on a two-dimensional transverse plane coordinate domain, which serves as height distribution information characterizing the three-dimensional morphology of the posterior pole surface of the eyeball.

4. The method for generating and standardizing the three-dimensional morphological parameters of the rear pole as described in claim 1, characterized in that, The step involves determining the rear pole analysis region in the analysis reference coordinate system and then meshing the rear pole analysis region to generate meshed surface data composed of multiple mesh nodes, including: S41, In the analysis reference coordinate system, the line connecting the center of the optic disc and the center of the macula is used as the reference baseline. Based on the preset extension range, the posterior pole analysis area covering the area around the optic disc and the area around the macula is determined. S42, on the two-dimensional transverse plane projection of the rear pole analysis region, a two-dimensional planar mesh is generated according to the preset mesh resolution; S43, Based on the height distribution information, determine the height value corresponding to each grid node in the two-dimensional planar grid; S44 combines the height value of each grid node with the two-dimensional planar grid to construct a grid node with a three-dimensional spatial position.

5. The method for generating and standardizing the three-dimensional morphological parameters of the rear pole as described in claim 1, characterized in that, The step of smoothing and filtering the gridded surface data, and then fitting and generating a reference surface representing the overall morphological trend of the posterior pole of the eyeball based on the smoothed and filtered gridded surface data, includes: S51, Spatial domain smoothing filtering is performed on the height values ​​of each grid node in the gridded surface data to eliminate high-frequency noise and local small fluctuations, so as to obtain smooth gridded surface data. S52, Based on the smooth gridded surface data, a low-order polynomial surface representing the overall bending trend of the posterior pole of the eyeball is obtained through a polynomial surface fitting algorithm. S53, the low-order polynomial surface is used as the reference surface to define a smooth reference surface that reflects the macroscopic anatomical curvature of the posterior pole in the analysis reference coordinate system.

6. The method for generating and standardizing the three-dimensional morphological parameters of the rear pole as described in claim 1, characterized in that, The calculation of the geometric features at each grid node within the post-polar analysis region based on the residual height value and the spatial distribution of grid nodes includes: S71, Based on the two-dimensional plane coordinates of each grid node in the gridded surface data and the corresponding residual height value, construct a residual height field function with two-dimensional plane coordinates as independent variables and residual height value as dependent variables; S72, on the residual height field function, based on the residual height values ​​of each grid node and its neighboring nodes, the first-order partial derivative and / or second-order partial derivative at each network node are calculated by numerical differentiation method. S73, calculate the slope at each network node based on the first-order partial derivative, and / or calculate the average curvature and / or Gaussian curvature at each network node based on the first-order and second-order partial derivatives, as geometric features.

7. The method for generating and standardizing the three-dimensional morphological parameters of the rear pole as described in claim 1, characterized in that, The parameterization is achieved by extracting quantitative indicators through statistical analysis and combining the quantitative indicators into a set of parameter vectors as morphological parameters of the posterior pole. The quantitative indicators include at least a global statistic reflecting the overall degree of concavity of the posterior pole, a spatial distribution difference reflecting morphological asymmetry, and a regional extreme value or gradient statistic reflecting the characteristics of local curvature changes.

8. The method for generating and standardizing the three-dimensional morphological parameters of the rear pole according to claim 1, characterized in that, The standardization of the rear electrode morphological parameters to generate a standardized set of rear electrode morphological parameters that is comparable across devices and time points includes: S91: Obtain the original dimensions and numerical range of each parameter in the morphological parameters of the rear pole; S92: According to the preset standardization rules, the morphological parameters of the rear pole are normalized in terms of scale. S93: The posterior pole morphological parameters after scale normalization are compared with the corresponding statistical distribution characteristics in the preset reference population database, and statistical standardization is performed to obtain standardized parameter values. S94: Combine the standardized parameter values ​​to generate the standardized polarity parameter set.

9. A system for generating and standardizing three-dimensional morphological parameters of the rear pole, characterized in that, The system is used to implement the method for generating and standardizing the three-dimensional morphological parameters of the rear pole as described in any one of claims 1 to 8, and the system includes: The image acquisition module is used to acquire wide-area optical coherence tomography image data of the subject's eyeball; The three-dimensional reconstruction module, connected to the image acquisition module, is used to perform three-dimensional volume data reconstruction on the wide-area optical coherence tomography image data to obtain height distribution information characterizing the three-dimensional morphology of the posterior pole surface of the eyeball. The coordinate system construction module, connected to the three-dimensional reconstruction module, is used to construct an analytical reference coordinate system for the posterior pole of the eyeball based on the spatial relationship between the center of the optic disc and the center of the macula. The meshing module, connected to the coordinate system construction module, is used to determine the rear pole analysis region under the analysis reference coordinate system and perform meshing processing on the rear pole analysis region to generate meshed surface data composed of multiple mesh nodes, wherein each mesh node is associated with a three-dimensional spatial position. The reference surface fitting module, connected to the meshing processing module, is used to perform smoothing filtering on the meshed surface data, and to fit and generate a reference surface representing the overall morphological trend of the posterior pole of the eyeball based on the smoothed meshed surface data. The residual calculation module, connected to the reference surface fitting module, is used to calculate the residual height value of each grid node. The residual height value is the difference between the original height value of the grid node and the corresponding height value of the reference surface at the grid node. The geometric feature calculation module, connected to the residual calculation module, is used to calculate the geometric features at each grid node in the post-polar analysis area based on the residual height value and the spatial distribution of grid nodes. The parametric characterization module, connected to the geometric feature calculation module, is used to parametrically characterize the three-dimensional structural morphology of the rear pole analysis area based on the geometric features and residual height value, and generate rear pole morphological parameters to describe the degree of rear pole concavity, morphological asymmetry and local curvature change characteristics. The parameter standardization module, connected to the parameterization characterization module, is used to standardize the morphological parameters of the rear pole to generate a standardized set of morphological parameters of the rear pole that is comparable across devices and time points.

10. A storage medium for computer-readable storage, wherein, The storage medium stores one or more programs, which can be executed by one or more processors to implement the method for generating and standardizing the three-dimensional morphological parameters of the rear pole as described in any one of claims 1 to 8.

Images